The majority of applications running over packet networks involve point-to-point working. Over the last few years, there has been growing interest in applications involving multiple participants and, increasingly, where these participants are all simultaneously involved in the communication. This interest has strengthened with the introduction of the MBone (the Internet multicasting backbone) and with the range of services made possible by ATM and those envisaged for Broadband ISDN. This thesis discusses the potential for a wide variety of multi-party applications. It examines their detailed requirements and the support mechanisms needed to meet these requirements. The work is presented as a dissertation and a collection of work published over a period of about ten years and as such draws together work on multi-party communication undertaken by the author and postgraduate students under her supervision. The major contribution of the thesis and the most recent work concerns multicast routing strategies capable of supporting high-bandwidth delay sensitive applications. A new heuristic is introduced which is shown to offer efficient routing solutions whilst ensuring that delays to each participant are kept within a bound. The heuristic is reasonably simple and is shown to perform well under a variety of conditions. The chapters of the these leading up to the work on multicast routing present the earlier published work. Architectural frameworks are presented which extend existing protocol reference models to offer multicast support mechanisms at appropriate hierarchical levels with a view to flexible yet efficient use of the network. One important support mechanism is group management and a system developed in the context of an integrated services network is described. This comprises a group management database together with a collection of flexible group management procedures capable of supporting a wide variety of applications.

Many problems in programming involve pairs of computations that cancel out each other's effects; some examples include parsing/printing, embed- ding/ projection, marshalling/unmarshalling, compressing/ de-com pressing etc. To avoid duplication of effort, the paradigm of bidirectional programming aims at to allow the programmer to write a single program that expresses both computations. Despite being a promising idea, existing studies mainly focus on the view-update problem in databases and its variants; and the impact of bidirectional programming has not reached the wider community. The goal of this thesis is to demonstrate, through concrete language designs and case studies, the relevance of bidirectional programming, in areas of computer science that have not been previously explored. In this thesis, we will argue for the importance of bidirectional programming in programming language design and compiler implementation. As evidence for this, we will propose a technique for incremental refactoring, which relies for its correctness on a bidirectional language and its properties, and devise a framework for implementing program transformations, with bidirectional properties that allow program analyses to be carried out in the transformed program, and have the results reported in the source program. Our applications of bidirectional programming to new areas bring up fresh challenges. This thesis also reflects on the challenges, and studies their impact to the design of bidirectional systems. We will review various design goals, including expressiveness, robustness, updatability, efficiency and easy of use, and show how certain choices, especially regarding updatability, can have significant influence on the effectiveness of bidirectional systems.

Graph drawing, or the automatic layout of graphs, is a challenging problem. There are several search-based methods for graph drawing that are based on optimising a fitness function which is formed from a weighted sum of multiple criteria. This thesis proposes a new neighbourhood search-based method that uses a tabu search coupled with path relinking in order to optimise such fitness functions for general graph layouts with undirected straight lines. None of these methods have been previously used in general multi-criteria graph drawing. Tabu search uses a memory list to speed up searching by avoiding previously tested solutions, while the path relinking method generates new solutions by exploring paths that connect high quality solutions. We use path relinking periodically within the tabu search procedure to speed up the identification of good solutions. We have evaluated our new method against the commonly used neighbourhood search optimisation techniques: hill climbing and simulated annealing. Our evaluation examines the quality of the graph layout (fitness function's value) and the speed of the layout in terms of the number of the evaluated solutions required to draw a graph. We also examine the relative scalability of our method. Our experimental results were applied to both random graphs and a real-world dataset. We show that our method outperforms both hill climbing and simulated annealing by producing a better layout in a lower number of evaluated solutions. In addition, we demonstrate that our method has greater scalability as it can lay out larger graphs than the state-of-the-art neighbourhood search-based methods. Finally, we show that similar results can be produced in a real world setting by testing our method against a standard public graph dataset.

This thesis examines the concept of graduateness in computing education. Graduateness is related to efforts to articulate the outcomes of a university education. It is commonly defined as the attributes all graduates should develop by the time they graduate regardless of university attended or discipline studied (Glover, Law and Youngman 2002). This work takes a different perspective grounded in disciplinary and institutional contexts. It aims to explore how graduates make sense of their experiences studying computing within their wider learning trajectories. The research presented here uses a narrative approach. Whilst narrative methodologies are not commonly used in computing education, people construct stories both to make sense of their experiences and to integrate the "past, present, and an anticipated future" (McAdams 1985, p.120). Stories are then a particularly appropriate way of examining the sense people make of their learning experiences. This work draws on narrative interviews with graduates from the School of Computing at the University of Kent and Olin College of Engineering in the United States. It contributes a new perspective about the effect of a computing education beyond short-term outcome measures and proposes several analytic constructs that expose significant aspects in participants' learning experiences. In this, it describes themes related to students' acquisition of disciplinary knowledge and examines the evolution of their stories of learning computing over time.

No description available.

Ant colony optimisation (ACO) is a metaheuristic to solve optimisation problems inspired by the foraging behaviour of ant colonies. It has been successfully applied to several types of optimisation problems, such as scheduling and routing, and more recently for the discovery of classification rules. The classification task in data mining aims at predicting the value of a given goal attribute for an example, based on the values of a set of predictor attributes for that example. Since real-world classification problems are generally described by nominal (categorical or discrete) and continuous (real-valued) attributes, classification algorithms are required to be able to cope with both nominal and continuous attributes. Current ACO classification algorithms have been designed with the limitation of discovering rules using nominal attributes describing the data. Furthermore, they also have the limitation of not coping with more complex types of classification problems e.g., hierarchical multi-label classification problems. This thesis investigates the extension of ACO classification algorithms to cope with the aforementioned limitations. Firstly, a method is proposed to extend the rule construction process of ACO classification algorithms to cope with continuous attributes directly. Four new ACO classification algorithms are presented, as well as a comparison between them and well-known classification algorithms from the literature. Secondly, an ACO classification algorithm for the hierarchical problem of protein function prediction which is a major type of bioinformatics problem addressed in this thesis is presented. Finally, three different approaches to extend ACO classification algorithms to the more complex case of hierarchical multi-label classification are described, elaborating on the ideas of the proposed hierarchical classification ACO algorithm. These algorithms are compare against state-of-the-art decision tree induction algorithms for hierarchical multi-label classification in the context of protein function prediction. The computational results of experiments with a wide range of data sets including challenging protein function prediction data sets with very large number.

No description available.

No description available.

Distributed computing systems, with a high degree of interaction, cooperation, and sharing of resources between large numbers of computing elements, are becoming critical to the working of many enterprises. The evolution of services and networks along with the development of interorganisational, distributed applications require distributed management techniques, systems and tools. Different platforms are currently available to assist the development of distributed applications, hiding the underlying diversity and physical distribution of different computers, operating systems and network protocols. However, they supply neither facilities for automatic management of the support services they provide to application developers nor of the applications themselves. A random allocation of software components to nodes and the lack of resource management policies may lead to poor performance by having some nodes overloaded while others are idle. This thesis proposes an approach to distributed systems management, addressing in particular the distribution of the workload submitted to a distributed system by its users. A management architecture based on the ODP Reference Model and the OSI Management Model is presented. Distributed System managers, each responsible for a management domain, and Node Managers interact to allocate services to suitable nodes, considering the services requirements and the resources available. A prototype implementation is described which demonstrates how the concepts and mechanisms that form the architecture can be realised. This, together with a qualitative evaluation, shows the benefits of incorporating this management approach in a distributed environment: service creation and distribution, as well as resource management, is made transparent to platform users; the workload submitted to the system is automatically distributed; and services are provided with the requirements they need.

Formal methods for the specification and development of distributed systems have traditionally supported a top-down, linear development process. However, there is a growing awareness in the (distributed) software engineering community that, in addition to a `vertical' structuring into design phases and steps, the specification and development of complex systems should also be structured `horizontally' according to, so called, viewpoints. Such multi-viewpoint development methods raise two important new issues though, viz. consistency and composition of viewpoint specifications. The aim of this thesis is to provide formal techniques to support multi-viewpoint specification and development. In particular, it deals with process algebraic techniques for viewpoint specification, consistency checking and viewpoint composition. An important issue is the formal definition of consistency, in particular when different viewpoints are expressed using different specification techniques -- a situation that is referred to as unbalanced (as opposed to balanced) consistency. To this end, a mathematical framework for formal, viewpoint oriented system development is elaborated by abstracting from specific viewpoint models and specific formal techniques. A number of existing process algebraic specification techniques based on, so called, implementation relations, are investigated on their suitability for viewpoint specification. The thesis contains a comprehensive study of all possible binary consistency relations between specifications in these techniques, resulting in a spectrum of consistency relations. It also provides techniques for composition in these specification techniques. In addition, a new mixed term specification technique is developed that combines the expressiveness of modal logics with the structuring capabilities of process calculi. Consistency and composition turn out to have precisely the desired properties, which makes it an ideal formalism for partial process specification. Moreover, because of its expressiveness, it can serve as a pivotal formalism for unbalanced consistency checking.